Bimodal Hand Vein Recognition System using Support Vector Machine

There are many types of biometric identifiers such as fingerprint, retina, or iris that can be used to recognize individuals. The hand veins are within the body which makes it difficult to forge and the vein pattern is unique for different individuals. Since there is a lack of study that considers both palm and dorsal vein on hands for recognition, further research on this area might be needed. This study was conducted to discover new ways to implement bimodal hand vein recognition. The system takes advantage of two NoIR cameras connected to a multi-camera adapter which was mounted on Raspberry Pi Model 3B+ in acquiring images of dorsal and palm vein exposed to infrared light. The quality of the acquired images was improved using the proposed image pre-processing techniques. Feature points were successfully obtained using the ORB algorithm. The results obtained showed that using Support Vector Machine as a classifier for hand veins was good and generated an accuracy of 97.16%.